Home | History | Annotate | Line # | Download | only in netinet
tcp_var.h revision 1.98
      1 /*	$NetBSD: tcp_var.h,v 1.98 2003/06/23 11:02:16 martin Exp $	*/
      2 
      3 /*
      4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
      5  * All rights reserved.
      6  *
      7  * Redistribution and use in source and binary forms, with or without
      8  * modification, are permitted provided that the following conditions
      9  * are met:
     10  * 1. Redistributions of source code must retain the above copyright
     11  *    notice, this list of conditions and the following disclaimer.
     12  * 2. Redistributions in binary form must reproduce the above copyright
     13  *    notice, this list of conditions and the following disclaimer in the
     14  *    documentation and/or other materials provided with the distribution.
     15  * 3. Neither the name of the project nor the names of its contributors
     16  *    may be used to endorse or promote products derived from this software
     17  *    without specific prior written permission.
     18  *
     19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     29  * SUCH DAMAGE.
     30  */
     31 
     32 /*
     33  *      @(#)COPYRIGHT   1.1 (NRL) 17 January 1995
     34  *
     35  * NRL grants permission for redistribution and use in source and binary
     36  * forms, with or without modification, of the software and documentation
     37  * created at NRL provided that the following conditions are met:
     38  *
     39  * 1. Redistributions of source code must retain the above copyright
     40  *    notice, this list of conditions and the following disclaimer.
     41  * 2. Redistributions in binary form must reproduce the above copyright
     42  *    notice, this list of conditions and the following disclaimer in the
     43  *    documentation and/or other materials provided with the distribution.
     44  * 3. All advertising materials mentioning features or use of this software
     45  *    must display the following acknowledgements:
     46  *      This product includes software developed by the University of
     47  *      California, Berkeley and its contributors.
     48  *      This product includes software developed at the Information
     49  *      Technology Division, US Naval Research Laboratory.
     50  * 4. Neither the name of the NRL nor the names of its contributors
     51  *    may be used to endorse or promote products derived from this software
     52  *    without specific prior written permission.
     53  *
     54  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
     55  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     56  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
     57  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
     58  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
     59  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
     60  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
     61  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
     62  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
     63  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
     64  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     65  *
     66  * The views and conclusions contained in the software and documentation
     67  * are those of the authors and should not be interpreted as representing
     68  * official policies, either expressed or implied, of the US Naval
     69  * Research Laboratory (NRL).
     70  */
     71 
     72 /*-
     73  * Copyright (c) 1997, 1998, 1999, 2001 The NetBSD Foundation, Inc.
     74  * All rights reserved.
     75  *
     76  * This code is derived from software contributed to The NetBSD Foundation
     77  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
     78  * NASA Ames Research Center.
     79  *
     80  * Redistribution and use in source and binary forms, with or without
     81  * modification, are permitted provided that the following conditions
     82  * are met:
     83  * 1. Redistributions of source code must retain the above copyright
     84  *    notice, this list of conditions and the following disclaimer.
     85  * 2. Redistributions in binary form must reproduce the above copyright
     86  *    notice, this list of conditions and the following disclaimer in the
     87  *    documentation and/or other materials provided with the distribution.
     88  * 3. All advertising materials mentioning features or use of this software
     89  *    must display the following acknowledgement:
     90  *	This product includes software developed by the NetBSD
     91  *	Foundation, Inc. and its contributors.
     92  * 4. Neither the name of The NetBSD Foundation nor the names of its
     93  *    contributors may be used to endorse or promote products derived
     94  *    from this software without specific prior written permission.
     95  *
     96  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     97  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     98  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     99  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
    100  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
    101  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
    102  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
    103  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
    104  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
    105  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
    106  * POSSIBILITY OF SUCH DAMAGE.
    107  */
    108 
    109 /*
    110  * Copyright (c) 1982, 1986, 1993, 1994, 1995
    111  *	The Regents of the University of California.  All rights reserved.
    112  *
    113  * Redistribution and use in source and binary forms, with or without
    114  * modification, are permitted provided that the following conditions
    115  * are met:
    116  * 1. Redistributions of source code must retain the above copyright
    117  *    notice, this list of conditions and the following disclaimer.
    118  * 2. Redistributions in binary form must reproduce the above copyright
    119  *    notice, this list of conditions and the following disclaimer in the
    120  *    documentation and/or other materials provided with the distribution.
    121  * 3. All advertising materials mentioning features or use of this software
    122  *    must display the following acknowledgement:
    123  *	This product includes software developed by the University of
    124  *	California, Berkeley and its contributors.
    125  * 4. Neither the name of the University nor the names of its contributors
    126  *    may be used to endorse or promote products derived from this software
    127  *    without specific prior written permission.
    128  *
    129  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
    130  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
    131  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
    132  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
    133  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
    134  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
    135  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
    136  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
    137  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
    138  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
    139  * SUCH DAMAGE.
    140  *
    141  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
    142  */
    143 
    144 #ifndef _NETINET_TCP_VAR_H_
    145 #define _NETINET_TCP_VAR_H_
    146 
    147 #if defined(_KERNEL_OPT)
    148 #include "opt_inet.h"
    149 #include "opt_mbuftrace.h"
    150 #endif
    151 
    152 /*
    153  * Kernel variables for tcp.
    154  */
    155 
    156 #include <sys/callout.h>
    157 
    158 /*
    159  * Tcp control block, one per tcp; fields:
    160  */
    161 struct tcpcb {
    162 	int	t_family;		/* address family on the wire */
    163 	struct ipqehead segq;		/* sequencing queue */
    164 	struct callout t_timer[TCPT_NTIMERS];/* tcp timers */
    165 	short	t_state;		/* state of this connection */
    166 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
    167 	uint32_t t_rxtcur;		/* current retransmit value */
    168 	short	t_dupacks;		/* consecutive dup acks recd */
    169 	u_short	t_peermss;		/* peer's maximum segment size */
    170 	u_short	t_ourmss;		/* our's maximum segment size */
    171 	u_short t_segsz;		/* current segment size in use */
    172 	char	t_force;		/* 1 if forcing out a byte */
    173 	u_int	t_flags;
    174 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
    175 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
    176 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
    177 #define	TF_NOOPT	0x0008		/* don't use tcp options */
    178 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
    179 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
    180 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
    181 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
    182 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
    183 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
    184 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
    185 #define	TF_CANT_TXSACK	0x1000		/* other side said I could not SACK */
    186 #define	TF_IGNR_RXSACK	0x2000		/* ignore received SACK blocks */
    187 #define	TF_REASSEMBLING	0x4000		/* we're busy reassembling */
    188 
    189 
    190 	struct	mbuf *t_template;	/* skeletal packet for transmit */
    191 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
    192 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
    193 	struct	callout t_delack_ch;	/* delayed ACK callout */
    194 /*
    195  * The following fields are used as in the protocol specification.
    196  * See RFC783, Dec. 1981, page 21.
    197  */
    198 /* send sequence variables */
    199 	tcp_seq	snd_una;		/* send unacknowledged */
    200 	tcp_seq	snd_nxt;		/* send next */
    201 	tcp_seq	snd_up;			/* send urgent pointer */
    202 	tcp_seq	snd_wl1;		/* window update seg seq number */
    203 	tcp_seq	snd_wl2;		/* window update seg ack number */
    204 	tcp_seq	iss;			/* initial send sequence number */
    205 	u_long	snd_wnd;		/* send window */
    206 	tcp_seq snd_recover;		/* for use in fast recovery */
    207 /* receive sequence variables */
    208 	u_long	rcv_wnd;		/* receive window */
    209 	tcp_seq	rcv_nxt;		/* receive next */
    210 	tcp_seq	rcv_up;			/* receive urgent pointer */
    211 	tcp_seq	irs;			/* initial receive sequence number */
    212 /*
    213  * Additional variables for this implementation.
    214  */
    215 /* receive variables */
    216 	tcp_seq	rcv_adv;		/* advertised window */
    217 /* retransmit variables */
    218 	tcp_seq	snd_max;		/* highest sequence number sent;
    219 					 * used to recognize retransmits
    220 					 */
    221 /* congestion control (for slow start, source quench, retransmit after loss) */
    222 	u_long	snd_cwnd;		/* congestion-controlled window */
    223 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
    224 					 * for slow start exponential to
    225 					 * linear switch
    226 					 */
    227 /*
    228  * transmit timing stuff.  See below for scale of srtt and rttvar.
    229  * "Variance" is actually smoothed difference.
    230  */
    231 	uint32_t t_rcvtime;		/* time last segment received */
    232 	uint32_t t_rtttime;		/* time we started measuring rtt */
    233 	tcp_seq	t_rtseq;		/* sequence number being timed */
    234 	int32_t	t_srtt;			/* smoothed round-trip time */
    235 	int32_t	t_rttvar;		/* variance in round-trip time */
    236 	uint32_t t_rttmin;		/* minimum rtt allowed */
    237 	u_long	max_sndwnd;		/* largest window peer has offered */
    238 
    239 /* out-of-band data */
    240 	char	t_oobflags;		/* have some */
    241 	char	t_iobc;			/* input character */
    242 #define	TCPOOB_HAVEDATA	0x01
    243 #define	TCPOOB_HADDATA	0x02
    244 	short	t_softerror;		/* possible error not yet reported */
    245 
    246 /* RFC 1323 variables */
    247 	u_char	snd_scale;		/* window scaling for send window */
    248 	u_char	rcv_scale;		/* window scaling for recv window */
    249 	u_char	request_r_scale;	/* pending window scaling */
    250 	u_char	requested_s_scale;
    251 	u_int32_t ts_recent;		/* timestamp echo data */
    252 	u_int32_t ts_recent_age;	/* when last updated */
    253 	u_int32_t ts_timebase;		/* our timebase */
    254 	tcp_seq	last_ack_sent;
    255 
    256 /* SACK stuff */
    257 	struct ipqehead timeq;		/* time sequenced queue (for SACK) */
    258 
    259 /* path MTU discovery blackhole detection */
    260 	int t_mtudisc;			/* perform mtudisc for this tcb */
    261 
    262 /* pointer for syn cache entries*/
    263 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
    264 };
    265 
    266 #ifdef _KERNEL
    267 /*
    268  * TCP reassembly queue locks.
    269  */
    270 static __inline int tcp_reass_lock_try __P((struct tcpcb *))
    271 	__attribute__((__unused__));
    272 static __inline void tcp_reass_unlock __P((struct tcpcb *))
    273 	__attribute__((__unused__));
    274 
    275 static __inline int
    276 tcp_reass_lock_try(tp)
    277 	struct tcpcb *tp;
    278 {
    279 	int s;
    280 
    281 	/*
    282 	 * Use splvm() -- we're blocking things that would cause
    283 	 * mbuf allocation.
    284 	 */
    285 	s = splvm();
    286 	if (tp->t_flags & TF_REASSEMBLING) {
    287 		splx(s);
    288 		return (0);
    289 	}
    290 	tp->t_flags |= TF_REASSEMBLING;
    291 	splx(s);
    292 	return (1);
    293 }
    294 
    295 static __inline void
    296 tcp_reass_unlock(tp)
    297 	struct tcpcb *tp;
    298 {
    299 	int s;
    300 
    301 	s = splvm();
    302 	tp->t_flags &= ~TF_REASSEMBLING;
    303 	splx(s);
    304 }
    305 
    306 #ifdef DIAGNOSTIC
    307 #define	TCP_REASS_LOCK(tp)						\
    308 do {									\
    309 	if (tcp_reass_lock_try(tp) == 0) {				\
    310 		printf("%s:%d: tcpcb %p reass already locked\n",	\
    311 		    __FILE__, __LINE__, tp);				\
    312 		panic("tcp_reass_lock");				\
    313 	}								\
    314 } while (/*CONSTCOND*/ 0)
    315 #define	TCP_REASS_LOCK_CHECK(tp)					\
    316 do {									\
    317 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
    318 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
    319 		    __FILE__, __LINE__, tp);				\
    320 		panic("tcp reass lock check");				\
    321 	}								\
    322 } while (/*CONSTCOND*/ 0)
    323 #else
    324 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
    325 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
    326 #endif
    327 
    328 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
    329 #endif /* _KERNEL */
    330 
    331 /*
    332  * Queue for delayed ACK processing.
    333  */
    334 #ifdef _KERNEL
    335 extern int tcp_delack_ticks;
    336 void	tcp_delack(void *);
    337 
    338 #define TCP_RESTART_DELACK(tp)						\
    339 	callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks,		\
    340 	    tcp_delack, tp)
    341 
    342 #define	TCP_SET_DELACK(tp)						\
    343 do {									\
    344 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
    345 		(tp)->t_flags |= TF_DELACK;				\
    346 		TCP_RESTART_DELACK(tp);					\
    347 	}								\
    348 } while (/*CONSTCOND*/0)
    349 
    350 #define	TCP_CLEAR_DELACK(tp)						\
    351 do {									\
    352 	if ((tp)->t_flags & TF_DELACK) {				\
    353 		(tp)->t_flags &= ~TF_DELACK;				\
    354 		callout_stop(&(tp)->t_delack_ch);			\
    355 	}								\
    356 } while (/*CONSTCOND*/0)
    357 #endif /* _KERNEL */
    358 
    359 /*
    360  * Compute the current timestamp for a connection.
    361  */
    362 #define	TCP_TIMESTAMP(tp)	(tcp_now - (tp)->ts_timebase)
    363 
    364 /*
    365  * Handy way of passing around TCP option info.
    366  */
    367 struct tcp_opt_info {
    368 	int		ts_present;
    369 	u_int32_t	ts_val;
    370 	u_int32_t	ts_ecr;
    371 	u_int16_t	maxseg;
    372 };
    373 
    374 /*
    375  * Data for the TCP compressed state engine.
    376  */
    377 union syn_cache_sa {
    378 	struct sockaddr sa;
    379 	struct sockaddr_in sin;
    380 #if 1 /*def INET6*/
    381 	struct sockaddr_in6 sin6;
    382 #endif
    383 };
    384 
    385 struct syn_cache {
    386 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
    387 	struct callout sc_timer;		/* rexmt timer */
    388 	union {					/* cached route */
    389 		struct route route4;
    390 #ifdef INET6
    391 		struct route_in6 route6;
    392 #endif
    393 	} sc_route_u;
    394 #define sc_route4	sc_route_u.route4
    395 #ifdef INET6
    396 #define sc_route6	sc_route_u.route6
    397 #endif
    398 	long sc_win;				/* advertised window */
    399 	int sc_bucketidx;			/* our bucket index */
    400 	u_int32_t sc_hash;
    401 	u_int32_t sc_timestamp;			/* timestamp from SYN */
    402 	u_int32_t sc_timebase;			/* our local timebase */
    403 	union syn_cache_sa sc_src;
    404 	union syn_cache_sa sc_dst;
    405 	tcp_seq sc_irs;
    406 	tcp_seq sc_iss;
    407 	u_int sc_rxtcur;			/* current rxt timeout */
    408 	u_int sc_rxttot;			/* total time spend on queues */
    409 	u_short sc_rxtshift;			/* for computing backoff */
    410 	u_short sc_flags;
    411 
    412 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
    413 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
    414 
    415 	struct mbuf *sc_ipopts;			/* IP options */
    416 	u_int16_t sc_peermaxseg;
    417 	u_int16_t sc_ourmaxseg;
    418 	u_int8_t sc_request_r_scale	: 4,
    419 		 sc_requested_s_scale	: 4;
    420 
    421 	struct tcpcb *sc_tp;			/* tcb for listening socket */
    422 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
    423 };
    424 
    425 struct syn_cache_head {
    426 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
    427 	u_short sch_length;			/* # entries in bucket */
    428 };
    429 
    430 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
    431 #ifdef INET6
    432 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
    433 #endif
    434 #ifndef INET6
    435 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
    436 #else
    437 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
    438 				? intotcpcb(sotoinpcb(so)) \
    439 				: in6totcpcb(sotoin6pcb(so)))
    440 #endif
    441 
    442 /*
    443  * The smoothed round-trip time and estimated variance
    444  * are stored as fixed point numbers scaled by the values below.
    445  * For convenience, these scales are also used in smoothing the average
    446  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
    447  * With these scales, srtt has 3 bits to the right of the binary point,
    448  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
    449  * binary point, and is smoothed with an ALPHA of 0.75.
    450  */
    451 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
    452 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
    453 
    454 /*
    455  * The initial retransmission should happen at rtt + 4 * rttvar.
    456  * Because of the way we do the smoothing, srtt and rttvar
    457  * will each average +1/2 tick of bias.  When we compute
    458  * the retransmit timer, we want 1/2 tick of rounding and
    459  * 1 extra tick because of +-1/2 tick uncertainty in the
    460  * firing of the timer.  The bias will give us exactly the
    461  * 1.5 tick we need.  But, because the bias is
    462  * statistical, we have to test that we don't drop below
    463  * the minimum feasible timer (which is 2 ticks).
    464  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
    465  * is the same as the multiplier for rttvar.
    466  */
    467 #define	TCP_REXMTVAL(tp) \
    468 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
    469 
    470 /*
    471  * Compute the initial window for slow start.
    472  */
    473 #define	TCP_INITIAL_WINDOW(iw, segsz) \
    474 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
    475 	 ((segsz) * (iw)))
    476 
    477 /*
    478  * TCP statistics.
    479  * Many of these should be kept per connection,
    480  * but that's inconvenient at the moment.
    481  */
    482 struct	tcpstat {
    483 	u_quad_t tcps_connattempt;	/* connections initiated */
    484 	u_quad_t tcps_accepts;		/* connections accepted */
    485 	u_quad_t tcps_connects;		/* connections established */
    486 	u_quad_t tcps_drops;		/* connections dropped */
    487 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
    488 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
    489 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
    490 	u_quad_t tcps_rttupdated;	/* times we succeeded */
    491 	u_quad_t tcps_delack;		/* delayed acks sent */
    492 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
    493 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
    494 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
    495 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
    496 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
    497 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
    498 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
    499 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
    500 					   shortage */
    501 	u_quad_t tcps_pmtublackhole;	/* PMTUD blackhole detected */
    502 
    503 	u_quad_t tcps_sndtotal;		/* total packets sent */
    504 	u_quad_t tcps_sndpack;		/* data packets sent */
    505 	u_quad_t tcps_sndbyte;		/* data bytes sent */
    506 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
    507 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
    508 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
    509 	u_quad_t tcps_sndprobe;		/* window probes sent */
    510 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
    511 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
    512 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
    513 
    514 	u_quad_t tcps_rcvtotal;		/* total packets received */
    515 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
    516 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
    517 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
    518 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
    519 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
    520 	u_quad_t tcps_rcvshort;		/* packets received too short */
    521 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
    522 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
    523 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
    524 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
    525 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
    526 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
    527 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
    528 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
    529 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
    530 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
    531 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
    532 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
    533 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
    534 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
    535 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
    536 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
    537 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
    538 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
    539 
    540 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
    541 	u_quad_t tcps_noport;		/* no socket on port */
    542 	u_quad_t tcps_badsyn;		/* received ack for which we have
    543 					   no SYN in compressed state */
    544 
    545 	/* These statistics deal with the SYN cache. */
    546 	u_quad_t tcps_sc_added;		/* # of entries added */
    547 	u_quad_t tcps_sc_completed;	/* # of connections completed */
    548 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
    549 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
    550 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
    551 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
    552 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
    553 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
    554 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
    555 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
    556 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
    557 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
    558 
    559 	u_quad_t tcps_selfquench;	/* # of ENOBUFS we get on output */
    560 };
    561 
    562 /*
    563  * Names for TCP sysctl objects.
    564  */
    565 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
    566 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
    567 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
    568 #define	TCPCTL_MSSDFLT		4	/* default seg size */
    569 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
    570 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
    571 #if 0	/*obsoleted*/
    572 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
    573 #endif
    574 #define	TCPCTL_INIT_WIN		8	/* initial window */
    575 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
    576 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
    577 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
    578 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
    579 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
    580 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
    581 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
    582 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
    583 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
    584 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
    585 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
    586 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
    587 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
    588 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
    589 #if 0	/*obsoleted*/
    590 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
    591 #endif
    592 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
    593 #define	TCPCTL_DELACK_TICKS	25	/* # ticks to delay ACK */
    594 #define	TCPCTL_INIT_WIN_LOCAL	26	/* initial window for local nets */
    595 #define	TCPCTL_IDENT		27	/* rfc 931 identd */
    596 #define	TCPCTL_MAXID		28
    597 
    598 #define	TCPCTL_NAMES { \
    599 	{ 0, 0 }, \
    600 	{ "rfc1323",	CTLTYPE_INT }, \
    601 	{ "sendspace",	CTLTYPE_INT }, \
    602 	{ "recvspace",	CTLTYPE_INT }, \
    603 	{ "mssdflt",	CTLTYPE_INT }, \
    604 	{ "syn_cache_limit", CTLTYPE_INT }, \
    605 	{ "syn_bucket_limit", CTLTYPE_INT }, \
    606 	{ 0, 0 },\
    607 	{ "init_win", CTLTYPE_INT }, \
    608 	{ "mss_ifmtu", CTLTYPE_INT }, \
    609 	{ "sack", CTLTYPE_INT }, \
    610 	{ "win_scale", CTLTYPE_INT }, \
    611 	{ "timestamps", CTLTYPE_INT }, \
    612 	{ "compat_42", CTLTYPE_INT }, \
    613 	{ "cwm", CTLTYPE_INT }, \
    614 	{ "cwm_burstsize", CTLTYPE_INT }, \
    615 	{ "ack_on_push", CTLTYPE_INT }, \
    616 	{ "keepidle",	CTLTYPE_INT }, \
    617 	{ "keepintvl",	CTLTYPE_INT }, \
    618 	{ "keepcnt",	CTLTYPE_INT }, \
    619 	{ "slowhz",	CTLTYPE_INT }, \
    620 	{ "newreno",	CTLTYPE_INT }, \
    621 	{ "log_refused",CTLTYPE_INT }, \
    622 	{ 0, 0 }, \
    623 	{ "rstppslimit", CTLTYPE_INT }, \
    624 	{ "delack_ticks", CTLTYPE_INT }, \
    625 	{ "init_win_local", CTLTYPE_INT }, \
    626 	{ "ident", CTLTYPE_STRUCT }, \
    627 }
    628 
    629 struct sysctl_tcp_ident_args {
    630 	struct in_addr raddr;
    631 	u_int rport;
    632 	struct in_addr laddr;
    633 	u_int lport;
    634 };
    635 
    636 #ifdef _KERNEL
    637 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
    638 #ifdef INET6
    639 extern	struct in6pcb tcb6;
    640 #endif
    641 extern	struct tcpstat tcpstat;	/* tcp statistics */
    642 extern	u_int32_t tcp_now;	/* for RFC 1323 timestamps */
    643 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
    644 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
    645 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
    646 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
    647 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
    648 extern	int tcp_mssdflt;	/* default seg size */
    649 extern	int tcp_init_win;	/* initial window */
    650 extern	int tcp_init_win_local;	/* initial window for local nets */
    651 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
    652 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
    653 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
    654 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
    655 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
    656 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
    657 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
    658 extern	int tcp_log_refused;	/* log refused connections */
    659 
    660 extern	int tcp_rst_ppslim;
    661 
    662 extern	int tcp_syn_cache_size;
    663 extern	struct syn_cache_head tcp_syn_cache[];
    664 extern	u_long syn_cache_count;
    665 
    666 #ifdef MBUFTRACE
    667 extern	struct mowner tcp_rx_mowner;
    668 extern	struct mowner tcp_tx_mowner;
    669 extern	struct mowner tcp_mowner;
    670 #endif
    671 
    672 #define	TCPCTL_VARIABLES { \
    673 	{ 0 },					\
    674 	{ 1, 0, &tcp_do_rfc1323 },		\
    675 	{ 1, 0, &tcp_sendspace },		\
    676 	{ 1, 0, &tcp_recvspace },		\
    677 	{ 1, 0, &tcp_mssdflt },			\
    678 	{ 1, 0, &tcp_syn_cache_limit },		\
    679 	{ 1, 0, &tcp_syn_bucket_limit },	\
    680 	{ 0 },					\
    681 	{ 1, 0, &tcp_init_win },		\
    682 	{ 1, 0, &tcp_mss_ifmtu },		\
    683 	{ 1, 0, &tcp_do_sack },			\
    684 	{ 1, 0, &tcp_do_win_scale },		\
    685 	{ 1, 0, &tcp_do_timestamps },		\
    686 	{ 1, 0, &tcp_compat_42 },		\
    687 	{ 1, 0, &tcp_cwm },			\
    688 	{ 1, 0, &tcp_cwm_burstsize },		\
    689 	{ 1, 0, &tcp_ack_on_push },		\
    690 	{ 1, 0, &tcp_keepidle },		\
    691 	{ 1, 0, &tcp_keepintvl },		\
    692 	{ 1, 0, &tcp_keepcnt },			\
    693 	{ 1, 1, 0, PR_SLOWHZ },			\
    694 	{ 1, 0, &tcp_do_newreno },		\
    695 	{ 1, 0, &tcp_log_refused },		\
    696 	{ 0 },					\
    697 	{ 1, 0, &tcp_rst_ppslim },		\
    698 	{ 1, 0, &tcp_delack_ticks },		\
    699 	{ 1, 0, &tcp_init_win_local },		\
    700 }
    701 
    702 #ifdef __NO_STRICT_ALIGNMENT
    703 #define	TCP_HDR_ALIGNED_P(th)	1
    704 #else
    705 #define	TCP_HDR_ALIGNED_P(th)	((((vaddr_t) (th)) & 3) == 0)
    706 #endif
    707 
    708 int	 tcp_attach __P((struct socket *));
    709 void	 tcp_canceltimers __P((struct tcpcb *));
    710 struct tcpcb *
    711 	 tcp_close __P((struct tcpcb *));
    712 #ifdef INET6
    713 void	 tcp6_ctlinput __P((int, struct sockaddr *, void *));
    714 #endif
    715 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
    716 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
    717 struct tcpcb *
    718 	 tcp_disconnect __P((struct tcpcb *));
    719 struct tcpcb *
    720 	 tcp_drop __P((struct tcpcb *, int));
    721 void	 tcp_dooptions __P((struct tcpcb *,
    722 	    u_char *, int, struct tcphdr *, struct tcp_opt_info *));
    723 void	 tcp_drain __P((void));
    724 #ifdef INET6
    725 void	 tcp6_drain __P((void));
    726 #endif
    727 void	 tcp_established __P((struct tcpcb *));
    728 void	 tcp_init __P((void));
    729 #ifdef INET6
    730 int	 tcp6_input __P((struct mbuf **, int *, int));
    731 #endif
    732 void	 tcp_input __P((struct mbuf *, ...));
    733 u_long	 tcp_mss_to_advertise __P((const struct ifnet *, int));
    734 void	 tcp_mss_from_peer __P((struct tcpcb *, int));
    735 struct tcpcb *
    736 	 tcp_newtcpcb __P((int, void *));
    737 void	 tcp_notify __P((struct inpcb *, int));
    738 #ifdef INET6
    739 void	 tcp6_notify __P((struct in6pcb *, int));
    740 #endif
    741 u_int	 tcp_optlen __P((struct tcpcb *));
    742 int	 tcp_output __P((struct tcpcb *));
    743 void	 tcp_pulloutofband __P((struct socket *,
    744 	    struct tcphdr *, struct mbuf *, int));
    745 void	 tcp_quench __P((struct inpcb *, int));
    746 #ifdef INET6
    747 void	 tcp6_quench __P((struct in6pcb *, int));
    748 #endif
    749 int	 tcp_reass __P((struct tcpcb *, struct tcphdr *, struct mbuf *, int *));
    750 int	 tcp_respond __P((struct tcpcb *, struct mbuf *, struct mbuf *,
    751 	    struct tcphdr *, tcp_seq, tcp_seq, int));
    752 void	 tcp_rmx_rtt __P((struct tcpcb *));
    753 void	 tcp_setpersist __P((struct tcpcb *));
    754 void	 tcp_slowtimo __P((void));
    755 struct mbuf *
    756 	 tcp_template __P((struct tcpcb *));
    757 void	 tcp_trace __P((int, int, struct tcpcb *, struct mbuf *, int));
    758 struct tcpcb *
    759 	 tcp_usrclosed __P((struct tcpcb *));
    760 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
    761 int	 tcp_usrreq __P((struct socket *,
    762 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
    763 void	 tcp_xmit_timer __P((struct tcpcb *, uint32_t));
    764 tcp_seq	 tcp_new_iss __P((struct tcpcb *, tcp_seq));
    765 tcp_seq  tcp_new_iss1 __P((void *, void *, u_int16_t, u_int16_t, size_t,
    766 	    tcp_seq));
    767 
    768 int	 syn_cache_add __P((struct sockaddr *, struct sockaddr *,
    769 		struct tcphdr *, unsigned int, struct socket *,
    770 		struct mbuf *, u_char *, int, struct tcp_opt_info *));
    771 void	 syn_cache_unreach __P((struct sockaddr *, struct sockaddr *,
    772 	   struct tcphdr *));
    773 struct socket *syn_cache_get __P((struct sockaddr *, struct sockaddr *,
    774 		struct tcphdr *, unsigned int, unsigned int,
    775 		struct socket *so, struct mbuf *));
    776 void	 syn_cache_init __P((void));
    777 void	 syn_cache_insert __P((struct syn_cache *, struct tcpcb *));
    778 struct syn_cache *syn_cache_lookup __P((struct sockaddr *, struct sockaddr *,
    779 		struct syn_cache_head **));
    780 void	 syn_cache_reset __P((struct sockaddr *, struct sockaddr *,
    781 		struct tcphdr *));
    782 int	 syn_cache_respond __P((struct syn_cache *, struct mbuf *));
    783 void	 syn_cache_timer __P((void *));
    784 void	 syn_cache_cleanup __P((struct tcpcb *));
    785 
    786 int	tcp_newreno __P((struct tcpcb *, struct tcphdr *));
    787 #endif
    788 
    789 #endif /* _NETINET_TCP_VAR_H_ */
    790